The correct answer is Strong consistency ensures that any reader in any region always sees the most recently committed version of an item.

With strong consistency Azure Cosmos DB provides linearizability so any read anywhere will reflect the latest committed write across the account. This guarantees that readers in different regions do not observe older versions once a write is committed. Keep in mind that strong consistency has higher latency and availability trade offs when you distribute data globally and it is not available if you enable multi master writes on the account.

Strong consistency requires that you commit to using Cosmos DB as your only storage solution is incorrect because consistency levels define read and write semantics and do not force you to use Cosmos DB as the sole storage system in your architecture.

Under strong consistency two readers in separate regions might read different versions of the same record at the same time is incorrect because strong consistency prevents that behavior and ensures readers see the most recently committed version.

Strong consistency implies that when two clients write the same item the write that arrives last always wins is incorrect because strong consistency defines read visibility and ordering guarantees and it does not by itself specify conflict resolution policies such as last writer wins in multi master scenarios.

Full AWS Practitioner Certification Question